Ra 0.8 Is Not a Complete Surface Specification: What Overseas Buyers Need to Compare

By Daoyu Guan

A surface-roughness value needs a parameter, a unit, an identified surface, and a defined measurement and acceptance context. “Ra 0.8” alone leaves too much unstated for comparing supplier capability or approving a component. Ra and Rz describe different aspects of a profile, and a unit conversion does not convert one parameter into the other. Keep the drawing's standard and edition visible, agree on the relevant measurement conditions, and distinguish a typical process result from an order-specific commitment. A supplier page can support preliminary selection without becoming a substitute for engineering agreement.

An overseas buyer asks for a smooth sealing face. The factory's English page lists “surface finish: Ra 0.8.” The buyer's drawing specifies a particular face, a different parameter, and conditions the website never mentions. Both parties may believe they are discussing surface finish while comparing different requirements.

For manufacturing GEO, the aim is to preserve the question behind the number: which surface must do what, under which agreed definition?

A profile number summarizes a particular measurement

Surface texture is richer than one height value. A profile measurement samples a trace across a surface and evaluates it under a specified method. The parameter summarizes an aspect of that assessed profile; it is not a complete description of how the component will function.

In familiar terms, Ra describes average absolute profile deviations, while Rz is sensitive to profile heights under the governing definition and evaluation procedure. Mitutoyo's surface-roughness guide illustrates profile filtering and several parameters in a document explicitly based on older DIN EN ISO references. It is useful for understanding those illustrations, but its old default tables and acceptance rules should not be silently applied to a different current specification.

Two surfaces can share an Ra value while having different peak, valley, spacing, or directional characteristics. That is a reason to ask what the drawing and function require, not a reason to replace Ra with a supposedly superior parameter in every application.

An arithmetic illustration makes the information loss visible. Take eight equally weighted heights in arbitrary units. The sequence −1, −1, −1, −1, 1, 1, 1, 1 has a mean height of zero and a mean absolute height of 1. So does −4, 0, 0, 0, 0, 0, 0, 4. Their peak-to-valley spans are 2 and 8. These are invented arrays, not measured profiles or standard-compliant Ra/Rz calculations. They simply show how the same average can conceal very different extremes.

Preserve the standard and edition

As reviewed for this article, ISO's listing for ISO 21920-2:2021 identifies terms, definitions, and surface-texture parameters and lists ISO 4287:1997 among its withdrawn predecessors. The page identifies a corrected version dated June 2022 and a revision stage. It still lists the 2021 edition as published.

ISO 21920-3:2021 addresses specification operators and lists ISO 4288:1996 as a withdrawn predecessor. These public listings establish the publications' status and broad scope; they do not provide a full paid standard or settle every drawing interpretation.

A newer published standard does not mean an editor should silently rewrite a customer's older drawing. Keep the invoked edition and ask the authorized engineering parties how the requirement is to be interpreted or changed. A product page that says only “ISO standard finish” is too broad to resolve that conversation.

Units and parameters are different dimensions of meaning

Micrometres and microinches are units of length. Ra and Rz identify parameters. Converting the unit preserves the parameter; it does not turn Ra into Rz.

Using the exact inch-to-metre relationship, 0.8 micrometres is approximately 31.50 microinches. Likewise, 32 microinches is 0.8128 micrometres. A rounded “32” and an exact “0.8” therefore require attention to the specified limit and rounding convention. The arithmetic alone does not decide acceptance.

There is no universal fixed multiplier that converts any surface's Ra into its Rz. A comparison chart may supply rough estimates for a limited purpose, but it cannot establish conformity to an order-specific Rz requirement from an Ra result. Ask for the required parameter measured under the agreed procedure.

These calculations are unit illustrations, not a recommendation to change a buyer's drawing or substitute a nominal equivalent.

The instrument mode matters

A measuring instrument can support several standards without making their settings interchangeable. Mitutoyo's SJ-220 product page lists roughness analysis under EN ISO 21920 and several other standards, including EN ISO 4287. The availability of those modes is a useful reminder to record the mode actually used.

For a supplier review, ask how the report identifies the standard, parameter, units, filters, evaluation length or other applicable settings, measurement location, and direction. The exact set and defaults must come from the governing specification and agreed procedure; this article does not prescribe one universal configuration.

Also distinguish instrument availability from a demonstrated measurement on the feature in question. A small bore, a short land, or a feature close to an edge may require a different assessment from an accessible flat coupon. A photograph of an instrument is evidence that it was photographed, not evidence that it can evaluate every claimed surface.

A result near the limit needs an agreed decision rule

Suppose an order sets an upper limit of Ra 0.80 micrometres for one identified surface. A hypothetical laboratory reports 0.79 micrometres with expanded uncertainty U = 0.04 micrometres and a stated coverage basis. The illustrative interval is 0.75 to 0.83 micrometres. Its upper endpoint exceeds the specified limit, even though the reported central value is below it.

Expanded uncertainty is an interval quantity. NIST's metrology glossary explains that interpreting its coverage depends on justified assumptions about the underlying distribution. Keep the reported uncertainty, coverage factor and stated coverage context together. The interval is not a guarantee that every possible value lies inside it, and this example supplies no probability that a real part is nonconforming.

For a conformity statement, the laboratory and customer need the applicable decision rule. Global ACI's June 2026 guidance, section 3.3, addresses defining the specification and rule and agreeing the selected rule with the customer unless it is already inherent in the requested specification or standard. Section 8 addresses compatibility with the applicable requirements, prior agreement and report documentation. This is laboratory conformity guidance, not a universal surface-texture acceptance instruction.

For arithmetic only, choose a binary rule with a guard band equal to U. The upper acceptance limit becomes 0.80 - 0.04 = 0.76 micrometres. The invented result of 0.79 would not be accepted under that particular rule. The guide illustrates this kind of rule in section 5.2.2; it also describes other approaches. The numbers do not establish which rule governs an actual drawing or contract, and they do not authorize adding a second guard band where the governing method already accounts for uncertainty.

Before approving the report, connect four records: the identified surface and requirement; the measurement method and result; the uncertainty and coverage statement; and the governing decision rule. Resolve a missing rule before converting a laboratory number into a delivery promise. A marketing editor should retain the agreed conclusion and its scope rather than manufacture a pass or fail from the headline value.

Locate the requirement on the actual component

“Surface finish” can refer to one sealing face, a sliding area, an aesthetic exterior, or a broad default drawing requirement. Those are different scopes. A single best result on one accessible face should not become a claim for the whole component.

For a hypothetical housing, identify face S1 as the controlled sealing surface. Keep its drawing reference and revision with the requirement. Record whether the proposed result concerns the machined condition, a later treated condition, or another agreed stage. Do not assume that a result measured before a later operation proves the final delivered condition.

If a buyer asks about a family of parts, explain which materials, geometries, and processes have actually been assessed. A typical process range can help early screening, provided it is clearly labeled as typical and does not promise every combination of material and geometry.

Write a claim that survives quotation

A defensible capability statement connects the process to its conditions. It might describe an established inspection method for a stated class of features and invite drawing review for the actual requirement. The manufacturer must supply and approve those facts.

Avoid publishing “all parts to Ra 0.8” because one sample achieved that value. Also avoid implying that tighter or smoother is always better. The correct requirement depends on function and the engineer's specification, not a marketing preference for the smallest number.

For Xindar's manufacturing content work, the editorial task is to retain the parameter, units, location, conditions, and approval boundary in English. Xindar's role as publisher or marketing provider does not establish metrology authority or approve the component.

Use a fictional comparison to find missing information

Supplier A offers “Ra 0.8 micrometres, typical on accessible ground faces, subject to drawing review.” Supplier B offers “Rz 4 micrometres” without a standard or measurement context. Supplier C shows a polished part and says “precision finish.” These are invented descriptions.

A is more explicit about some boundaries, but that does not prove it meets the buyer's actual requirement. B provides a different parameter that cannot be equated with A's result by an unsupported ratio. C supplies visual information without a measurable acceptance basis.

The next step is to send the same controlled requirement and ask each supplier to identify what it can assess, how it would measure the feature, and what remains conditional. Rank the completeness of those responses separately from the engineering decision about feasibility.

This is how educational content can improve supplier comparison: it gives the buyer a better question rather than manufacturing a winner from incomplete data.

Keep public text aligned with the approved record

Check every place the claim appears: the capability page, product table, brochure, quotation template, and any comparison article. A careful sentence can become misleading when shortened to a card reading “Ra 0.8 guaranteed.”

Preserve units during translation and formatting. A table heading can supply a unit, but that context may be lost if a row is copied elsewhere. Put the parameter and unit close enough to remain interpretable in an excerpt, while avoiding repetitive claims that expand the scope.

If an AI answer turns a conditional range into an unconditional promise, retain the answer and cited source, then locate the actual mismatch. Correct the source page where necessary. Do not assume that changing the page immediately changes every answer or that structured data can repair an unsupported technical claim.

Questions from engineering and marketing teams

Can Ra be converted into Rz?

Not by a universal fixed ratio valid for every surface and specification. They summarize different properties. Use the parameter required by the drawing and an agreed measurement procedure.

Is Ra 0.8 the same as 32 microinches?

They are close rounded descriptions, not identical exact values. Ra 0.8 micrometres converts to about 31.50 microinches; 32 microinches converts to 0.8128 micrometres. The governing limit and rounding rules determine whether the distinction matters for the order.

Must every old drawing be changed to ISO 21920?

This article does not establish that. Preserve the invoked requirement and resolve interpretation with the authorized engineering parties. A publisher should not silently substitute standards or acceptance rules.

Does a measured Ra of 0.79 automatically pass a 0.80 limit?

The two numbers alone do not settle conformity. The defined surface, applicable measurement procedure, uncertainty and governing decision rule matter. The invented guard-band example above explains one possible rule; it does not impose that rule on a real order.

Does a lower roughness value prove a better component?

It proves only the result supported by the measurement and its conditions. Functional suitability depends on the design and application. Appearance, roughness, waviness, geometry, and other requirements should not be collapsed into one marketing adjective.

Sources and method

The ISO public pages support publication identity, status, and broad scope; full standards were not purchased or reproduced. The Mitutoyo PDF is explicitly treated as a legacy explanatory source, and the current product page supports the existence of multiple analysis modes. NIST supports the uncertainty terminology; the June 2026 Global ACI guide supplies laboratory decision-rule context. All supplier comparisons and numerical acceptance examples are hypothetical. This article does not establish the applicable contract rule, calculate a shipment risk, or provide inspection or product approval.